A direct drive motor adaptive multi-modal operation control method

Abstract

The application discloses a kind of direct drive motor adaptive multi-modal operation control method, belong to motor control technical field.The method obtains motor operating state data by physical sensor network, after FPGA synchronous acquisition, in turn data pre-processing and feature extraction, obtain direct drive motor mode characteristics and store.Construction is based on decision gradient decision tree direct drive motor mode identification model, according to direct drive motor mode characteristics identification direct drive motor motor operating mode, automatically switch corresponding control strategy, while monitoring the operating performance index of direct drive motor under control strategy, judge whether to optimize control strategy internal parameter.The method realizes the adaptive control of direct drive motor multi-modal operation, improves control precision and system stability, with higher engineering application value.

Description

Technical Field

[0001] This invention relates to the field of motor control technology, and specifically to an adaptive multimodal operation control method for a direct-drive motor. Background Technology

[0002] Direct-drive motors are widely used in industrial automation fields such as new energy equipment and high-end manufacturing equipment. Their operation often involves switching between multiple modes, and the requirements for dynamic response, parameter matching, and system stability of control strategies vary significantly depending on the mode. In existing technologies, traditional control methods use a single strategy with fixed parameters, which makes it difficult to automatically switch to the optimal control strategy according to the real-time mode. Furthermore, the accuracy of mode recognition is low due to the single dimension of feature extraction. Parameter adjustment mostly relies on manual or offline optimization and cannot be dynamically optimized based on real-time performance indicators. At the same time, the lack of parameter transition logic during multi-mode switching can easily lead to problems such as torque pulsation. These shortcomings restrict the application of direct-drive motors in high-precision and high-dynamic-response scenarios.

[0003] Therefore, how to achieve adaptive control of multi-modal operation of direct-drive motors and improve control accuracy and system stability is a technical challenge that urgently needs to be solved in the field of motor control technology. To address this, an adaptive multi-modal operation control method for direct-drive motors is proposed. Summary of the Invention

[0004] The purpose of this invention is to provide an adaptive multimodal operation control method for direct drive motors to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides the following technical solution:

[0006] An adaptive multimodal operation control method for a direct-drive motor includes the following steps:

[0007] S1. Obtain the operating status data of the direct drive motor;

[0008] S2. Perform data preprocessing on the operating status data to obtain motor status data, extract features from the motor status data to obtain direct drive motor mode features, and store them in the database;

[0009] S3. Construct a direct drive motor pattern recognition model, input the direct drive motor pattern features into the direct drive motor pattern recognition model, and output the motor operating mode;

[0010] S4. Based on the motor's operating mode, the direct drive motor automatically switches and executes the corresponding control strategy. At the same time, it monitors the operating performance indicators of the direct drive motor under the control strategy, evaluates the operating performance indicators, and determines whether to optimize the internal parameters of the control strategy.

[0011] A preferred method for obtaining the operating status data of a direct-drive motor:

[0012] The operating status data of the direct drive motor is acquired through a physical sensor network. A field-programmable gate array (FPGA) is used as the control unit. A high-precision synchronous sampling trigger signal is generated by the internal clock management unit of the FPGA. This signal is transmitted to the multi-channel analog-to-digital converter (ADC) via a differential driver, driving the ADC to synchronously acquire the operating status data at a sampling frequency of not less than 10kHz. The FPGA and the multi-channel ADC achieve precise matching of sampling timing through a clock synchronization mechanism based on low-voltage differential signals. The clock synchronization error is controlled within ±50ns to ensure timing consistency.

[0013] The physical sensor network consists of a Hall current sensor, an isolation voltage sensor, a rotary transformer, a MEMS accelerometer, a torque sensor, and a temperature sensor. The physical sensor network is connected to a signal conditioning circuit via a shielded cable. The conditioning circuit includes anti-aliasing filtering, signal amplification, and bias adjustment modules to ensure that the signal amplitude input to the ADC is within the range of 0 to 5V.

[0014] Preferred method for preprocessing operational status data:

[0015] The outlier values ​​in the operating status data are detected and deleted using the interquartile range method, resulting in outlier-detected operating status data. The missing values ​​in the outlier-detected operating status data are then filled using linear interpolation to obtain the motor status data.

[0016] The interquartile range method is a statistical indicator used to measure the dispersion of data and is a method for identifying outliers in data.

[0017] The linear interpolation method is a method for linearly estimating an unknown intermediate point by using the known values ​​of two points.

[0018] The motor status data includes the three-phase instantaneous current I. a I b and I c DC bus voltage V dc Rotor angle θ, triaxial vibration acceleration a x a y and a z Output shaft torque T and motor winding temperature T ω .

[0019] A preferred method is to extract features from motor state data to obtain the features of the direct-drive motor mode:

[0020] Substituting the rotor angle θ into the formula for the rate of change of rotational speed, the triaxial vibration acceleration a... x a y and a zSubstituting into the vibration energy density formula, the three-phase instantaneous current I a I b and I c Substitute the copper loss power ratio formula and power factor formula into the motor winding temperature T. ω Substituting the formula for the rate of temperature rise into the calculation, the rate of change of rotational speed is obtained. Vibrational energy density E vib Copper loss power ratio P cu-ra Temperature rise rate α T and power factor PF;

[0021] The formula for the rate of change of rotational speed is:

[0022]

[0023] Where, θ t Let θ be the rotor angle at the t-th time. t-1 θ is the rotor angle at the (t-1)th time. t-2 Δt is the rotor angle at the (t-2)th time, and Δt is the sampling interval time;

[0024] The formula for the vibration energy density is:

[0025]

[0026] Where T1 is the time window;

[0027] The formula for the copper loss power ratio is:

[0028]

[0029] Among them, P to R is the total input power of the direct drive motor, and R is the resistance of each phase winding of the direct drive motor.

[0030] The power factor formula is:

[0031]

[0032] Among them, I rms This is the effective value of the line current of the direct drive motor, V. rms This is the effective value of the line current voltage of the direct drive motor, V. a V b and V c It is the three-phase instantaneous voltage, and T1 is the time window;

[0033] The formula for the temperature rise rate is:

[0034]

[0035] Where Δt is the sampling interval time, ΔT ωIt is the difference in the temperature of the motor windings between two adjacent samplings;

[0036] The Clark-Park transformation is performed on the three-phase instantaneous currents to convert them into q-axis currents I in the dq coordinate system. q and d-axis current I d Based on this, the q-axis current I is obtained. q The average q-axis current is obtained by calculating using the average current formula. Based on this, the average q-axis current is... Substituting the output shaft torque T into the torque ripple coefficient formula, we obtain the torque ripple coefficient K. rip ;

[0037] The formula for the average current is:

[0038]

[0039] Where T1 is the time window, It is the average q-axis current;

[0040] The formula for the torque ripple coefficient is:

[0041]

[0042] Where, k t It is the torque constant, T max and T min These are the maximum and minimum values ​​of the output shaft torque T;

[0043] For three-phase instantaneous current I a I b and I c The signal is converted to the frequency domain using Fast Fourier Transform (FFT) to obtain the effective values ​​I1, I2, ..., I of each harmonic component. n The current harmonic distortion rate (THD) is calculated using the current harmonic distortion rate formula. I Based on this, the rate of change of rotational speed is obtained. q-axis current mean Current Harmonic Distortion Rate (THD) I Vibration energy density E vib Copper loss power ratio P cu-ra Temperature rise rate α T Torque ripple coefficient K rip Characteristics of direct-drive motors with power factor (PF);

[0044] The formula for the current harmonic distortion rate is:

[0045]

[0046] Where I1 is the fundamental effective value.

[0047] Preferred method for constructing a pattern recognition model for direct-drive motors:

[0048] A feature set is constructed by obtaining several historical direct-drive motor mode features from the database. The feature set is divided into a training set and a test set in an 8:2 ratio. The motor operating modes are manually labeled on the training set and the test set. The training set is input into the decision gradient boosting tree for training, and the test set is input into the trained decision gradient boosting tree to obtain the predicted motor operating modes. The number of prediction error samples that are different from the predicted motor operating modes of the manually labeled test set is counted. If the number of prediction error samples accounts for more than 3% of the number of samples in the test set, the learning rate, tree depth, minimum number of leaf nodes, and number of iterations of the decision gradient boosting tree are readjusted and set, and the decision gradient boosting tree is trained. Otherwise, the trained decision gradient boosting tree is obtained, which is the direct-drive motor pattern recognition model. The motor operating modes include start-up mode, steady-state operation mode, speed regulation mode, and braking mode.

[0049] The decision gradient boosting decision tree is an ensemble learning algorithm that iteratively trains multiple decision trees, gradually fitting the residuals of the data, and finally weights and combines the prediction results of all trees to form a strong prediction model.

[0050] Preferably, the automatic switching of the direct drive motor adopts a smooth switching strategy, specifically including bumpless switching or hybrid switching. In bumpless switching, before the physical controller switches, the initial control output of the control strategy to be executed by the physical controller is initialized to the initial control output of the currently executed control strategy.

[0051] During the hybrid switching process, within a set transition time, the initial control output u of the control strategy currently being executed by the physical controller is adjusted according to the weighting coefficient α: 1→0. o The initial control output u of the control strategy to be switched n We perform weighted fusion to obtain the switching control quantity u, and then execute the output. The calculation of the switching control quantity satisfies: u = α·u o +(1-α)·u n The transition time ranges from 10ms to 100ms.

[0052] Preferably, the corresponding control strategy is executed by the physical controller of the direct drive motor, wherein the control strategy includes a ramp soft start strategy corresponding to the start mode, a vector control strategy corresponding to the steady-state operation mode, an adaptive fuzzy PID control strategy corresponding to the speed regulation mode, and a feedback braking control strategy corresponding to the braking mode.

[0053] The ramp soft start strategy gradually increases the motor input voltage and current through a linear or exponential ramp function to avoid the inrush current at the moment of startup and achieve a smooth increase in the speed of the direct drive motor. Its essence is to balance the contradiction between starting torque and mechanical shock by controlling the energy input rate.

[0054] The vector control strategy decomposes the three-phase current into independent excitation and torque components through coordinate transformation, thereby achieving decoupled control of speed and torque.

[0055] The adaptive fuzzy PID control strategy is based on a fuzzy inference system and dynamically adjusts the PID parameters according to the speed error and the rate of change of the error.

[0056] The regenerative braking control strategy converts the kinetic energy of the direct drive motor into electrical energy and feeds it back to the power grid. By controlling the q-axis current to be negative, braking torque is generated, while maintaining the DC bus voltage stability, thus achieving the dual goals of energy recovery and rapid stopping.

[0057] A preferred method is to evaluate operational performance indicators to determine whether the internal parameters of the control strategy need to be optimized.

[0058] The operating performance indicators include current surge coefficient, speed accuracy, speed range ratio, and DC bus voltage fluctuation.

[0059] If the current surge coefficient of the direct-drive motor in startup mode is greater than 2, the internal parameters of the ramp soft-start strategy are optimized using a particle swarm optimization algorithm. These internal parameters include the voltage ramp time T. r and current limiting value I lim The current impact coefficient is the ratio of the maximum starting current of the direct drive motor in starting mode to the rated current of the direct drive motor.

[0060] If the speed accuracy of the direct-drive motor is less than 99.5% in steady-state operation, the internal parameters of the vector control strategy are optimized using a particle swarm optimization algorithm. These internal parameters include the speed loop PID parameters. and and current loop PID parameters and Rotational speed accuracy is ω ref ω is the rated speed of the direct drive motor under steady-state operation. a The measured speed of the direct drive motor in steady-state operation mode;

[0061] If the speed range ratio of the direct-drive motor in speed regulation mode is less than 1:100, the internal parameters of the adaptive fuzzy PID control strategy are optimized using a particle swarm optimization algorithm. The internal parameters of the adaptive fuzzy PID control strategy include the initial values ​​of the PID parameters. and And the parameter adjustment sensitivity coefficient, the speed range ratio is the ratio of the maximum stable speed and the minimum stable speed of the direct drive motor in speed regulation mode;

[0062] If the DC bus voltage fluctuation of the direct-drive motor exceeds 5% in braking mode, the internal parameters of the regenerative braking control strategy are optimized using a particle swarm optimization algorithm. These internal parameters include the braking current reference value. and bus voltage threshold V dc-ref DC bus voltage fluctuation is V dc-max and V dc-min These represent the maximum and minimum DC bus voltages of the direct drive motor in braking mode.

[0063] Compared with the prior art, the present invention has the following beneficial effects:

[0064] 1. This invention achieves innovation in multimodal adaptive control technology system. By using physical sensor network and FPGA synchronous acquisition technology, a multi-dimensional feature set containing 8 core features is constructed. Combined with decision gradient boosting decision tree model, high-precision recognition of motor mode is achieved. Based on the recognition results, control strategy is automatically switched. With the help of disturbance-free switching or hybrid switching mechanism, the problems of single mode recognition dimension and large torque ripple during strategy switching in the prior art are solved, and adaptive matching of control strategy for direct drive motor under all working conditions is achieved.

[0065] 2. This invention improves the control performance and engineering application value of direct-drive motors. By dynamically optimizing the internal parameters of each control strategy through particle swarm optimization, the starting current impact coefficient is no greater than 2, the steady-state speed accuracy is no less than 99.5%, the speed range ratio is no greater than 1:100, and the DC bus voltage fluctuation is no greater than 5%. Compared with traditional fixed parameter control, it improves overall energy efficiency and reduces the cost of manual parameter adjustment. It effectively solves the problems of low control accuracy and reliance on manual parameter optimization in existing technologies, and is suitable for high-precision dynamic response scenarios such as new energy equipment. Attached Figure Description

[0066] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0067] Figure 1 This is a flowchart of the method steps of the present invention. Detailed Implementation

[0068] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be described in detail below. Obviously, the described embodiments are merely some embodiments of this invention, and not all embodiments. Based on the embodiments of this invention, all other implementation methods obtained by those skilled in the art without creative effort are within the scope of protection of this invention.

[0069] Examples, such as Figure 1 As shown, an adaptive multimodal operation control method for a direct-drive motor includes the following steps:

[0070] S001. Obtain the operating status data of the direct drive motor;

[0071] S002. Perform data preprocessing on the operating status data to obtain motor status data, extract features from the motor status data to obtain direct drive motor mode features, and store them in the database;

[0072] S003. Construct a direct drive motor pattern recognition model, input the direct drive motor pattern features into the direct drive motor pattern recognition model, and output the motor operating mode;

[0073] S004. Based on the motor's operating mode, the direct drive motor automatically switches and executes the corresponding control strategy. At the same time, it monitors the operating performance indicators of the direct drive motor under the control strategy, evaluates the operating performance indicators, and determines whether to optimize the internal parameters of the control strategy.

[0074] Furthermore, the working principle of the present invention will be illustrated below through embodiments:

[0075] Taking a 1.5MW direct-drive wind turbine as an example, the parameters of the direct-drive wind turbine are as follows: rated power 1.5MW, rated voltage 690V, rated speed 1200r / min, number of pole pairs 30, winding resistance 0.012Ω. It is equipped with a physical sensor network, FPGA synchronous acquisition unit, data processing server and motor controller. The FPGA adopts Xilinx Kintex-series, the ADC sampling frequency is set to 20kHz, and the clock synchronization error is controlled within ±30ns.

[0076] Hall current sensors are used to collect three-phase current, isolation voltage sensors monitor DC bus voltage, 16-bit rotary transformers obtain rotor position, MEMS accelerometers collect triaxial vibration signals, torque sensors measure output shaft torque, and PT100 temperature sensors monitor winding temperature. During the wind turbine startup process, the FPGA generates a synchronous sampling trigger signal to drive an 8-channel ADC to synchronously collect operating status data. Taking the startup scenario where the wind speed increases from 8 m / s to 12 m / s as an example, the peak value of the three-phase instantaneous current is 1200A, the initial value of the DC bus voltage is 500V, and the rotor angle change rate is 15° / ms.

[0077] Three-phase instantaneous currents from 1000 sets of operational status data collected during the startup phase were processed using interquartile range (IQR) analysis. 23 outliers were detected, exceeding the 1.5 IQR range. Five missing values ​​were filled using linear interpolation. After processing, the effective value of the three-phase instantaneous current was 950A, and the DC bus voltage stabilized at 650V. Based on this, feature extraction was performed to obtain the direct-drive motor mode characteristics. Taking speed change rate, vibration energy density, and current harmonic distortion rate as examples, the speed change rate formula was used... Taking the sampling interval Δt as 0.001s, the rotor angle θ at time t... t θ is 30°, at time t-1 t-1 θ is 15°, at time t-2 t-2 Given a radius of 0°, the calculated rate of change of rotational speed is 15000° / s. 2 Within a 100ms time window, the root mean square value of the triaxial vibration acceleration is 0.5g, and E is calculated. vib 5g 2 / s, perform FFT transformation on the three-phase instantaneous current, where the fundamental effective value is 900A, and calculate the THD. I It is 5.8%.

[0078] 10,000 historical direct-drive motor mode features were collected from the database and divided into a training set of 8,000 groups and a test set of 2,000 groups in an 8:2 ratio. 2,500 groups were manually labeled for start-up mode, 4,000 groups for steady-state operation, 2,000 groups for speed regulation, and 1,500 groups for braking. Model training was then performed using a decision gradient boosting decision tree with the following parameters: learning rate 0.1, tree depth 6, minimum number of leaf nodes 5, and number of iterations 100. After training, the test set identification results were as follows: for start-up mode, the actual test set had 500 samples, with 492 correctly predicted samples, resulting in an accuracy rate of 98.4%. For steady-state operation, the actual test set... The actual sample size was 800, with 790 correctly predicted samples, resulting in an accuracy rate of 98.8%. In speed regulation mode, the actual test set had 400 samples, with 391 correctly predicted samples, resulting in an accuracy rate of 97.8%. In braking mode, the actual test set had 300 samples, with 294 correctly predicted samples, resulting in an accuracy rate of 98.0%. The error rate of the initial training test set was 2.3%, or 46 / 2000, which is less than 3%, so retraining is not required. This yielded the direct drive motor pattern recognition model. The direct drive motor pattern features were then input into the direct drive motor pattern recognition model for processing to obtain the motor operating mode.

[0079] Taking the motor operating mode in speed regulation mode as an example, in a speed regulation scenario where the wind speed decreases from 12m / s to 10m / s, a hybrid switching strategy is adopted. The transition time is set to 50ms, and the weighting coefficient α decreases linearly from 1 to 0. During the switching process, the torque ripple coefficient decreases from 0.12 to 0.08, and the current fluctuation amplitude is less than 10%. In addition, in the optimization of the internal parameters of the control strategy, taking the start-up mode, steady-state operation, and braking mode as examples, the current impact coefficient at the initial start-up in the start-up mode is 2.8, which exceeds the threshold of 2. By optimizing the voltage ramp time from 1.5s to 2.2s and the current limit value from 1200A to 1000A through the particle swarm optimization algorithm, the current impact coefficient is reduced to 1.8. The measured speed accuracy under steady-state operation is 99.2%, which is lower than 99.5%. The speed loop PID parameters are optimized. It increased from 0.8 to 1.2. The speed accuracy was improved to 99.6% after the value was increased from 0.3 to 0.5. The DC bus voltage fluctuation was 7% in braking mode, which exceeded 5%. The braking current reference value was adjusted from -600A to -700A, and the bus voltage threshold was reduced from 700V to 680V. After optimization, the voltage fluctuation was reduced to 4.2%.

[0080] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them; modifications to the technical solutions described in the foregoing embodiments, or equivalent substitutions of some of the technical features, do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of this application, and should all be included within the protection scope of this application.

Claims

1. A method for adaptive multimodal operation control of a direct-drive motor, characterized in that, Includes the following steps: S1. Obtain the operating status data of the direct drive motor; S2. Perform data preprocessing on the operating status data to obtain motor status data, extract features from the motor status data to obtain direct drive motor mode features, and store them in the database; S3. Construct a direct drive motor pattern recognition model, input the direct drive motor pattern features into the direct drive motor pattern recognition model, and output the motor operating mode; S4. Based on the motor's operating mode, the direct drive motor automatically switches and executes the corresponding control strategy. At the same time, it monitors the operating performance indicators of the direct drive motor under the control strategy, evaluates the operating performance indicators, and determines whether to optimize the internal parameters of the control strategy. The method for obtaining the operating status data of the direct drive motor: The operating status data of the direct drive motor is acquired through a physical sensor network. The FPGA is used as a control unit to generate a synchronous sampling trigger signal to drive the multi-channel ADC to synchronously acquire the operating status data. The FPGA and the multi-channel ADC achieve precise matching of sampling timing through a clock synchronization mechanism. The physical sensor network consists of Hall current sensors, isolation voltage sensors, rotary transformers, MEMS accelerometers, torque sensors, and temperature sensors. The method for constructing a pattern recognition model for a direct-drive motor: A feature set is constructed by retrieving several historical direct-drive motor pattern features from the database. This feature set is then divided into a training set and a test set in an 8:2 ratio. The motor operating modes are manually labeled on both the training and test sets. The training set is then input into a decision gradient boosting tree for training, and the test set is input into the trained decision gradient boosting tree to obtain the predicted motor operating modes. The number of incorrect prediction samples that differ from the manually labeled motor operating modes in the test set is counted. If the number of incorrect prediction samples exceeds 3% of the total number of samples in the test set, the decision gradient boosting tree is retrained. Otherwise, the trained decision gradient boosting tree, i.e., the direct-drive motor pattern recognition model, is obtained.

2. The adaptive multimodal operation control method for a direct-drive motor according to claim 1, characterized in that, The method for preprocessing operational status data: The outlier values ​​in the operating status data are detected and deleted using the interquartile range method, resulting in outlier-detected operating status data. The missing values ​​in the outlier-detected operating status data are then filled using linear interpolation to obtain the motor status data. The motor status data includes three-phase instantaneous current, DC bus initial voltage, rotor angle, triaxial vibration acceleration, output shaft torque, and motor winding temperature.

3. The adaptive multimodal operation control method for a direct-drive motor according to claim 2, characterized in that, The method for extracting features from motor state data to obtain direct drive motor mode features: The rotor angle is substituted into the formula for the rate of change of rotational speed, the triaxial vibration acceleration is substituted into the formula for the vibration energy density, the three-phase instantaneous current is substituted into the formulas for the copper loss power ratio and the power factor, and the motor winding temperature is substituted into the formula for the rate of temperature rise. The calculations are then performed to obtain the rate of change of rotational speed, vibration energy density, copper loss power ratio, rate of temperature rise, and power factor. The Clark-Park transformation is performed on the three-phase instantaneous current to obtain the q-axis current. The average q-axis current is then calculated using the current mean formula. Based on this, the average q-axis current and the output shaft torque are substituted into the torque ripple coefficient formula to calculate the torque ripple coefficient. The effective values ​​of each harmonic component are obtained by performing a fast Fourier transform on the three-phase instantaneous current, and the current harmonic distortion rate is obtained by calculating it using the current harmonic distortion rate formula. Based on this, the characteristics of the direct drive motor mode, including the speed change rate, the average q-axis current, the current harmonic distortion rate, the vibration energy density, the copper loss power ratio, the temperature rise rate, the torque ripple coefficient, and the power factor, are obtained.

4. The adaptive multimodal operation control method for a direct-drive motor according to claim 3, characterized in that, The motor operating modes include starting mode, steady-state operation mode, speed regulation mode, and braking mode.

5. The adaptive multimodal operation control method for a direct-drive motor according to claim 1, characterized in that, The automatic switching of the direct drive motor adopts a smooth switching strategy, which can be either bumpless switching or hybrid switching.

6. The adaptive multimodal operation control method for a direct-drive motor according to claim 1, characterized in that, The corresponding control strategies are executed by the physical controller of the direct drive motor. The control strategies include a ramp soft start strategy for the start-up mode, a vector control strategy for the steady-state operation mode, an adaptive fuzzy PID control strategy for the speed regulation mode, and a feedback braking control strategy for the braking mode.

7. The adaptive multimodal operation control method for a direct-drive motor according to claim 1, characterized in that, The method for evaluating operational performance indicators and determining whether to optimize the internal parameters of the control strategy: The operating performance indicators include current surge coefficient, speed accuracy, speed range ratio, and DC bus voltage fluctuation. If the current surge coefficient of the direct drive motor is greater than 2 in the start-up mode, the internal parameters of the ramp soft start strategy are optimized using the particle swarm optimization algorithm. If the speed accuracy of the direct drive motor is less than 99.5% in steady-state operation mode, the internal parameters of the vector control strategy are optimized using the particle swarm optimization algorithm. If the speed range ratio of the direct drive motor in speed regulation mode is less than 1:100, the internal parameters of the adaptive fuzzy PID control strategy are optimized by particle swarm optimization algorithm. If the DC bus voltage fluctuation of the direct drive motor in braking mode is greater than 5%, the internal parameters of the regenerative braking control strategy will be optimized using a particle swarm optimization algorithm.

Images